Re: Proposal for isolation characters in Unicode and the unicode-bidi:isolate and unicode-bidi:plaintext definitions

>
>
> UBA-2 would thus not only add these characters, but require that
> conformant applications support the whole set of bidi format characters
> fully. That way, your database could specify that it should be viewed with
> a browser supporting UBA-2, and all would be fine.


It would be fine only in the CYA sense. That is, the simple-minded app
using out database has no idea about Unicode generally or UBA-2
specifically, and would simply output whatever it got from the database.
And the user would use whatever browser has been installed on his or her
machine for the past nine years and "works just fine". So whether the stuff
is declared as UBA-2 or not, the result is pretty much the same.


> PS: I still believe having two models for isolate, i.e. a different one
> for CSS and Unicode, is a recipe for trouble.


I totally agree. That is why I am suggesting the CSS spec changes. Once the
new Unicode characters are accepted and a standard UBA implementation (e.g.
ICU) supports them, a browser can implement the (updated) CSS spec by
passing the new characters to the new UBA implementation. A browser
implementing the updated CSS spec before the updated UBA implementation is
available will have a harder time, but they are not exactly having an easy
time implementing the current CSS spec for isolate anyway.

Aharon

On Tue, May 15, 2012 at 8:20 PM, Asmus Freytag <asmusf@ix.netcom.com> wrote:

> In order for your Database to rely on this, you would need to know that
> whatever display tool you are using supports the new characters. The only
> way to surface this would be to create a new label (UBA-2) which would be a
> shorthand for compliance that includes these characters.
>
> The current Unicode compliance model allows implementers to arbitrarily
> subset the supported characters. This is fine for scripts, because there
> may be other reasons (lack of fonts, lack of rendering engines) why some
> implementation is unable to deal with, say Arabic, or Syriac.
>
> But it makes the situation totally unpredictable if applications are able
> to freely subset which Bidi FORMAT characters they support.
>
> UBA-2 would thus not only add these characters, but require that
> conformant applications support the whole set of bidi format characters
> fully. That way, your database could specify that it should be viewed with
> a browser supporting UBA-2, and all would be fine.
>
> A./
>
> PS: I still believe having two models for isolate, i.e. a different one
> for CSS and Unicode, is a recipe for trouble.
>
>